LAB mode for post processing: drawbacks?

I often see various articles and tutorials recommending to switch to LAB for this or that operation. Sharpening comes to mind, and the latest blog post at lensrentals.com also indicates that LAB is superior when it comes to pushing colors. I'm wondering if there are any advantages to stay in RGB mode, and if so, for what type of things.

There have been raging debates about moving in and out of colorspaces, whether it be RGB>LAB or CMYK, etc. Some purport to demonstrate that the integrity of the data suffers during conversion, others respond by demonstrating that you can convert back and forth multiple times without any noticeable degradation in the file, etc.

There are certain things that are easier, more effective, or more efficient in LAB when compared to attempting them in RGB. If you need to perform a particular move in LAB, then convert, do what you need to do and convert back to RG B when you are finished.

I use LAB for various reasons, and even when I stay in RGB, I visualize tone and color in LAB (in the Photoshop info palette, for example). LAB is a much more intuitive way of expressing and analyzing tone and color. The trick is to be able to relate RGB changes in LAB and vice versa. If you think in LAB you can often determine what needs to be done to correct color but perform the corrections in RGB without having to convert. For example, you are performing color correction on an image with very few reference neutrals, but there is a clear view of the midday blue sky with clouds. We know clouds are essentially neutral and we know that the sky is blue with, maybe, a tiny bit of green (blue+green = cyan). If you use a color sampler and drop one on the cloud and one on the sky, you may find that they read too green (i.e., too negative in the A channel). You can correct this with RGB curves (pull down the green at the specific tone on the curve, add some red, etc.) so that the A channel numbers are correct for the sky and the A and B channels are close to 0 in the clouds. So, you are thinking and correcting according to LAB values, but doing it in RGB, avoiding the conversion.

If you want to increase the color contrast by pulling in the curves in A and B, just because you read someone's technique for doing this, then try it. Most of the time, the conversion itself isn't going to destroy the image, but the curves adjustment to increase saturation and color contrast may be taken too far. Remember that you will ultimately have to convert back to RGB, so if you push colors out of your target color space gamut while in LAB (which is easy to do) you will make a mess when you convert back to RGB.

So, the conversion back and forth between RGB and LAB is typically, in and of itself, not inherently bad - however, if you do things in LAB without understanding their consequences in terms of the conversion back to RGB, you can cause problems. Know your target colorspace and its limitations when making bold moves in LAB.

kirk

That was very informative, thanks!

See any of the books and tutorials from Dan Margulis for more detailed information. With great power comes great responsibility....

kirk

kirkt wrote in post #16201134
See any of the books and tutorials from Dan Margulis for more detailed information. With great power comes great responsibility....

kirk

+1 The Canyon Conundrum is my favorite Margulis book, but they're all good.

On lab: Nice read here: http://www.dgrin.com/s​howthread.php?t=18203